Do Democracies Support Violent Non-governmental Organizations Less Than Autocracies Do?

Abstract

This study draws attention to the question: Do democracies fight indirectly through material support to violent non-governmental organizations (VNGOs) which wage intrastate war against other states in general and against democracies in particular, or are democracies less warlike by proxy? The main conclusions are: democracies are less warlike by proxy than non-democratic states, and the more democratic the regime the lower the probability that it will support VNGOs waging war against other states, both in general and against other democracies in particular. The results do not unequivocally support both the monadic or dyadic argument of democratic peaceful behaviour

Notes

1. I use violent non-governmental organizations (VNGOs) instead of terror organisations for two reasons: first the term ‘terror’ has strong political connotations, especially when we come to decide which government is supporting terrorism and which is not. Second, the data-set I use does not collect data on terror organizations/guerrilla movements/insurgent movements etc., but rather on external support in intrastate wars, whether to the central government or to the non-governmental organization which waged war against it. I do not use the term Non-state Armed Groups either, because these entities are not groups that might be random gatherings or primary groups. Instead, I use the term organization, which is the more accurate sociological term.

2. The databases on terror attacks document thousands of VNGOs which fit their definition of terror organizations. See Global Terrorism Database (GTD), ITERATE, RAND Database of Worldwide Terrorism Incidents.

3. For the claim about conventional war: (Ember, Ember, and Russett 1992) and for support for governments in intrastate wars (Goldman and Abulof 2016).

4. For descriptive statistics see Table 1 in the e-appendix https://drive.google.com/open?id = 0B3_UERJhnQn-TzA2dTkyY0RDVWs.

5. In order to compare the effects of different variables, Bs in each model were multiplied by the variable’s mean value. The supporters’ regime type is graded as 5–8 among eight variables (depending on the measure polity or civil liberties, and whether it is a dyadic or monadic variable), which was found to be significantly connected to the support of VNGOs, Table 2 in the e-appendix https://drive.google.com/open?id=0B3_UERJhnQn-TzA2dTkyY0RDVWs.